Скачать презентацию I acknowledge the generous support from the Human Скачать презентацию I acknowledge the generous support from the Human

f26e478449e32051a251f89facd30d42.ppt

  • Количество слайдов: 23

I acknowledge the generous support from the Human and Social Dynamics of Change program, I acknowledge the generous support from the Human and Social Dynamics of Change program, National Science Foundation (Decision, Risk, and Management Sciences division)—award # 0756539 Collaborators on this project include: Scott Ferson and Jack Siegrist (Applied Biomathematics, Setauket NY) Winston Harrington, Sandra Hoffman, and Elena Safirova (Resources for the Future) Paul Slovic and Branden Johnson (Decision Research, Eugene OR) Vicky Salin (Texas A&M) Dale Hattis (Clark University) Carl Cranor (Univ. of California, Riverside)

Outputs of this Portfolio of Research: • Article documenting inattention to uncertainty in regulatory Outputs of this Portfolio of Research: • Article documenting inattention to uncertainty in regulatory cost analysis, esp. as compared to advances in QUA for risk assessment (Penn Law); • Article proposing decision rules that can increase demand for high-quality information on costs and risks (Penn Law); • Development of a decision paradigm for “solution-focused risk assessment” (Penn Law); • Case study of (household) interindividual variability in regulatory cost– vehiclemiles traveled tax (Resources for the Future); • Case study of (manufacturer) interindividual variability in reg’y cost– safer poultry processing (Texas A&M/RFF); • Explorations of the inefficiencies created by imbalance across the risk/cost divide (Applied Biomathematics); • Article on “pitfalls of combining good data with bad” (AB);

Outputs of this Portfolio of Research (continued): • Methods for “advanced bias correction” in Outputs of this Portfolio of Research (continued): • Methods for “advanced bias correction” in cost-benefit analysis (AB); • September 2013 book entitled Does Regulation Kill Jobs? (Penn Law and chapter authors); • Extensive surveys (750 laypeople; 100 SRA members) that are the first steps towards a theory of regulatory “cost perception” (Penn Law/ Decision Research) • Article contrasting “default” assumptions in regulatory economics and QRA– reasons for inattention to uncertainty in economics (Clark Univ/Penn Law); • Article on the ethical questions raised by inattention to interindividual variability in regulatory cost burden (U-Cal Riverside; Penn Law);

Definitions of Cost and Benefit, and elaborations (spillovers) There are many coherent ways to Definitions of Cost and Benefit, and elaborations (spillovers) There are many coherent ways to parse costs and benefits: for example, • “the costs of pollution, ” “the benefits of agriculture, ” the “social cost of carbon”… • “costs are all the negative things that happen, benefits are all the positive things (basically the way the various regulatory “reform” bills in the 1990 s defined them) But these (and others) don’t focus attention on interventions (policies, regulations…), and don’t isolate the work done by risk scientists and by regulatory economists. For this project, we define benefits as all the changes (positives net of negatives) in goods not traded in markets (e. g. , longevity, quality-of-life, visibility); we define costs as all the changes in goods traded in markets (labor, capital, etc. ) “Benefits are what we buy with interventions: costs are what we give up to buy them. ” This puts regulatory economists on the “cost side” of the ledger, and risk scientists on the “risk side, ” along with economists who “valuate” benefits into monetary terms. No system isolates effects perfectly: in this one, what about a pollution-reduction policy that unclogs the arteries of workers, making them more productive and changing the supply curve of what they produce? That’s both a benefit and a negative cost.

“Transferring to Regulatory Economics the Risk-Analysis Approaches to Uncertainty, Interindividual Variability, and Other Phenomena”: “Transferring to Regulatory Economics the Risk-Analysis Approaches to Uncertainty, Interindividual Variability, and Other Phenomena”: • Uncertainty • Interindividual variability (a. k. a. “incidence” on the cost side) • Valuation (what is the function converting individual risk to monetized harm, especially at the de minimus end and the “intolerable” end? And (believe it or not) what is the function converting economic harm to monetized harm? • Equity (once the risk and cost valuation functions have been drawn, do we add everyone’s risk-harm and cost-harm equally, or weight them somehow? )

Risks and Costs: The Virtues of Balance: (in the intensity and rigor of analysis, Risks and Costs: The Virtues of Balance: (in the intensity and rigor of analysis, • in the clarity and honesty of presentation, • in the “individualization” of results, • in the valuation of effects in terms of individual and social utility, • in the analysis of counterveiling (risk-risk/cost-cost) effects, • in the efforts to understand lay and expert cognition, • in the interest in post hoc corroboration, etc. )

The Current Landscape of “Balance”: • Economists cajoling risk assessors to make their results The Current Landscape of “Balance”: • Economists cajoling risk assessors to make their results fit into the economic paradigm (e. g. , abandon the Rf. D/Rf. C/MOE measures); • Not a lot of improvement on the cost side (focus of this talk); • Occasional influential observations that (for example) it’s also not necessary for risk assessors to include information on uncertainty (!), because “many crucial economic policy decisions are made on the basis of point estimates of the gross domestic product, the unemployment rate, or the costs of major welfare or health care reform legislation, for example, without mathematical or even narrative descriptions of the considerable uncertainties” (Pres/Cong’l Commission on Risk Assessment and Management, 1996). So if point estimates are good enough for economists…

ERIC– pass the baton at this point to my esteemed mentor Hattis… ERIC– pass the baton at this point to my esteemed mentor Hattis…

How Uncertain could a cost estimate be? ? (cases from Harrington, Morgenstern, and Nelson, How Uncertain could a cost estimate be? ? (cases from Harrington, Morgenstern, and Nelson, RFF) Rule Ex Ante Est. Ex Post Analysis EPA 1986 ban on dinoseb $70 million - $400, 000 EPA 1988 aldicarb ban $135 million > $100 million SO 2 permits (Burtraw reanalysis) $250/permit $125/permit EPA 1999 reformul. gas 2. 3 ¢/gal 1. 8 - 2. 3 ¢/gal OSHA vinyl chloride (1974) $100 million (for $20 million for 1 ppm a less stringent level than 1 ppm) OSHA cotton dust (1978) $280 million $85 million OSHA Pb (1978) $150 million $20 million OSHA powered platforms (1989) - $1. 7 million - $600, 000 CARB LEV $170/car $83/car NOx trading in CA $11, 250/permit $1800/permit OSHA coke oven 3 specific firms would spend $91 million The 3 firms spent $5 -7 million

HIERARCHY OF HOW AN RIA COULD DEAL WITH UNCERTAINTY IN REGULATORY COST 1. Point HIERARCHY OF HOW AN RIA COULD DEAL WITH UNCERTAINTY IN REGULATORY COST 1. Point estimate of total cost (TC) only, with no mention of the word “uncertainty. ” 2. Point estimate of TC, but with narrative caveat about its uncertainty 3. Two or more point estimates of TC, because the RIA considers two or more different regulatory interventions. 4. Two or more point estimates of TC, because the RIA considers two or more mutually exclusive scenarios about how the regulated entities will comply with the regulation. 5. A “quasi-range” for TC (it looks like a range, but in fact is composed of the multiple point estimates from level 3 or 4). 6. A statement of uncertainty in TC derived from a rule of thumb (e. g. , “the point estimate plus or minus 10 percent”) 7. A statement of uncertainty in TC derived by fitting different estimates to a range or distribution. 8. A partial Monte Carlo analysis of uncertainty in TC (in general, combining assumed uncertainties in only a couple of key input variables). 9. An elaborate Monte Carlo analysis of parameter uncertainty in TC 10. An elaborate Monte Carlo analysis of uncertainty in TC, plus multiple models for estimating costs, with or without expert elicitation of the weights that could be applied to the probability of each model being correct.

REANALYSIS OF 75 EPA Rules from 1990 s by Hahn and Dudley: (single point REANALYSIS OF 75 EPA Rules from 1990 s by Hahn and Dudley: (single point estimate, no caveat(s)) (a “quasi-range” consisting of multiple pt. estimates) (two or more pt. estimates corresponding to different regulatory outcomes) (Monte Carlo simulation considering some of the sources of cost uncertainty)

1986(!) (Proposed EPA rule to establish treatment standards for hazardous wastes under RCRA– included 1986(!) (Proposed EPA rule to establish treatment standards for hazardous wastes under RCRA– included a fate and transport model, with a Monte Carlo procedure used, to back-calculate the concentrations of various substances in leachate that would be expected to exceed health-based limits in groundwater and surface water used for drinking– and also used Monte Carlo procedures to compare (via their CDFs ) different treatment technologies).

OSHA Chromium (VI) Final Rule (2006)– Cost Estimation: • 1 value for VSL ($6. OSHA Chromium (VI) Final Rule (2006)– Cost Estimation: • 1 value for VSL ($6. 8 M); • 1 value for own-price elasticity of demand (1. 0); • 2 discount rates (3%, 7%) “$282, 385, 793” (where are the pennies? ) Note- in OSHA’s 1997 MC rule, didn’t trust the # of furniture strippers found in Yellow Pages, so instead used [(total volume of MC used for this)/volume an “average” firm uses]– with no uncertainty.

From Krupnick et al. (2006): Not a Sure Thing: Making Regulatory Choices Under Uncertainty From Krupnick et al. (2006): Not a Sure Thing: Making Regulatory Choices Under Uncertainty (a “mock briefing” to senior EPA officials, but #s based closely on CAIR rule)

Even EPA’s Most Recent Major CBA… (The Benefits and Costs of the Clean Air Even EPA’s Most Recent Major CBA… (The Benefits and Costs of the Clean Air Act from 1990 to 2020, 238 pp. , April 2011) Monetized Annual Benefits in 2020 (in 2006 $): • 5 th percentile estimate: $250 billion • central estimate: $ 2 trillion • 95 th percentile estimate: $5. 7 trillion Monetized Costs: • $ 65 billion: “the degree of uncertainty cannot be reliably quantified… thus, we are unable to present specific high and low cost estimates. ” (although later (p. 3 -14, EPA does posit a 10% “technological learning rate, ” and lets it vary between 5% and 20%-- this creates a range of uncertainty in total cost of from $62 to $68 billion) Tables of Uncertainties: • • Table 3 -4 (cost uncertainties): 12 factors, 11 of them deemed to be “probably minor” Table 7 -6 (“potentially major sources of uncertainty for estimating the costs and benefits of the CAA”): 13 factors– 10 scientific, 3 valuation, none in cost.

FDA, Good Manufacturing Practices for Dietary Supplements Final Rule, June 25, 2007 Monte Carlo FDA, Good Manufacturing Practices for Dietary Supplements Final Rule, June 25, 2007 Monte Carlo simulation of both B and C, although no distributions used, only discrete (equiprobable? ) alternative estimates; for example: Benefits VSL ($3 M, $5 M, $7 M) Fraction of illnesses reported (1%, 2. 5%, 10%) Rule reduces recalls by (80%, 100%) Costs cost per test ($20, $60, $100) # establishments (1300, 1460) # batches per estab (+/- 50% of survey result)

POSSIBLE REASONS WHY REGULATORY ECONOMISTS MIGHT IGNORE (DEEMPHASIZE) UNCERTAINTY IN TOTAL COST Practical: • POSSIBLE REASONS WHY REGULATORY ECONOMISTS MIGHT IGNORE (DEEMPHASIZE) UNCERTAINTY IN TOTAL COST Practical: • Belief that uncertainty in cost is small and symmetric about the point (“best”? ) estimate used • Belief that cost uncertainty is small relative to risk uncertainty (the “cup of coffee or the national debt” problem in risk), so why bother…? • Belief that costs are significantly uncertain only in the long-term • The “unknowns” paradox (hard to estimate uncertainty for the most uncertain components, e. g, future technologies) Normative: • Belief that expected value provides all needed information (society is risk-neutral with respect to TC) • Belief that “EV plus 2 sigma” provides all needed info (Scott Farrow in “What Does an Economist Want”: “we want sufficient precision to distinguish positive from negative net benefit values. ”) Of course, this begs the question of why ask for all the uncertainty analysis in risk if “all you need to know” is that NB>0 with pr >. 95 • Strategic behavior has caused confusion: emphasis on “getting risk assessors to give us expected values instead of upper bounds” has economists holding fast to their own point estimates • Belief that costs are among the “social factors, ” or that they are “too important to quantify”

Bureaucratic: • Agency lawyers are intolerant of uncertainty in cost, and force economists to Bureaucratic: • Agency lawyers are intolerant of uncertainty in cost, and force economists to be overconfident, whereas they are more used to accepting risk estimates as uncertain, like it or not (or perhaps they feel ill-equipped to challenge the science and hence defer)? • Agency economists rarely rise to managerial positions, and so can’t impel change that would comport with their professional norms (p 29 of Williams) Disciplinary: • Economists may be less ingrained to emphasize error bounds (and may treat “defaults” as insiders’ information) • Inputs to cost estimates often come from agency engineers, who think in point estimates • Economists (or those asking them for point estimates) know that the current cost procedures produce “conservative” estimates and don’t want (the public) to see “all points left” [just as, perhaps, those calling for more analysis of interindividual variability in exposure thought the point estimates were “conservative” and wanted light shone on the rest of the distribution] • Are there in fact card-carrying (academic) economists analyzing regulatory cost? (there certainly are academic scientists in agencies working on risk assessment, but perhaps the economists are assigned to valuation and other work) • Rise of “Freakonomics” as ticket to fame in some of the field– finding clever relationships, not doing the “grunt work” of careful analysis of something as dull as “what it costs”